CRITICAL RESPONSE: I

Fall 1998
Volume 25, Number 1

Why is it that the most sustained encounter with Manet ever written--Michael Fried's Manet's Modernism, or, The Face of Painting in the 1860s--gives only one and a half pages (out of over six hundred) to A Bar at the Folies-Bergère?1 Basically, it's because the painting is not pivotal for Fried: it truncates the ternary relation between model, painting, and beholder that structures Manet's most characteristic works of the 1860s.2 In effect, A Bar at the Folies-Bergère lowers the pressure, by giving away things that had been held in expressive tension. And there's another reason, one not so much entangled with the book's purpose, namely, that the painting hasn't been considered crucial until recently. As Thierry de Duve points out, it is only since T. J. Clark's The Painting of Modern Life (1985) that the Bar has been central to Manet studies.3 So what does the painting have that makes it so irresistible? Why is it on the verge of becoming that bookend that matches the Déjeuner sur l'herbe, bracketing Manet and defining his generation, and even modernism itself? Why is it "one of the most pertinent challenges to art history today" (p. 000)? To get at these questions I am going to have to make a detour through a subject that may first seem more a matter of duty than a pleasure. But it isn't irrelevant; in fact, it is the key to the painting's popularity. The subject is geometry.

Geometry often seems beside the point, as if it were separate from painting's deeper questions almost by definition, and yet often enough there it is, at the very center of our interest. So I will begin with the unavoidable, irredeemable facts of geometry; then pull myself up, by slow stages, to a point where it will look as if I have left it behind; and then, if all goes well, it will become apparent that the annoying so-called trivia of lines and angles have been at stake all along. If you are uninterested in technical details, and incurious about your uninterst, please leap ahead to section 3, where I take up the "more interesting" argument about the popularity of the Bar.

I. Geometry and Nothing but Geometry

It seems that the Bar is "in perspective"--a very odd phrase, crucial to any account, which I consider to be almost perfectly opaque--because it has several perspectival objects in it. Each one can be doubted, and de Duve does so by counting up the errors in William Conger's diagram. But where perspective is concerned, errors and evidence are treacherous ground, and it is not difficult to shift the terms of explanation a little and show that each of de Duve's errors is itself mistaken.

First, the fruit bowl: Conger draws two foreshortened rectangles around its ellipses and locates a vanishing point (figure 2, number 4). De Duve finds this persuasive, but it isn't, because only two of its seventeen lines are determined by the painting.4 It is a reasonable guess, eyeballed from the look of the rest of the painting; in other words, it is made plausable by the objects that it supposedly helps explain--a typical situation in perspective.

Conger reads a number of such details and finds several distinct vanishing points. For de Duve, this amounts to overinterpretation because Manet's painting is done "in an offhand manner," and it would be brash to conclude that it has so many seperate vanishing points. Surely Manet could not have planted all those little "clues" simply in order ro reveal an arcane lack of self-consistency. On the contrary: "the artist has in fact laid down clues pointing to optics and even prompting a strictly Albertian perpectival reading of the painting" (p. 000). And that judgement leads de Duve to his central theme, the inconsistency between the figure of the barmaid and the reflections.

Already something extremely strange--by which I mean logically dubious--is happening to the argument. Conger makes a few errors, but he reveals enough to show that it's not going to be easy to talk about the painting's perpective. From this de Duve concludes that the Bar is self-contradictory, but not in the way that the forms themselves indicate. Instead we are to read the individual constructions as "clues" that Manet wanted us to think about "optics," and even "a stricly Albertian reading." That is itself isn't logical (under what circumstances should a chaos of unreliable, competing details be understood as evidence of the painter's desire to demonstrate a strong unity?). Furthermore, it then turns out that the only optics we need to concern ourselves with is that of the barmaid, her reflection, and the reflected customer.

These changes of subject (perspective to optics) and simplifications (many pieces of evidence to just three) are the signs of an argument gathering itself to go in a new direction. It is as if there has to be a way to talk about reflections and to make an optical proof without attending to all the optical evidence, or without attending to it all with equal rigor. For de Duve, "there must be a trick somewhere" if "the laws of optics can be shown to apply to an image that so obviously disregards them" (p.000). But why one trick? Why not several, or several dozen? (and why tricks? And why clues?) As long as the only inconsistency that counts, the only one pointed to by the optics, is one that occurs between the three figures, then it stands to reason that the painting is an either / or proposition. And that is why de Duve can say, with a very strict logic, "either the viewer or the mirror moves" (p.000).

Before I look at the consequences of this reduction, let me spend a moment on what it omits. According to de Duve's interpretation, Conger's multiple viewpoints cannot be taken literally because they would mean the painting isn't in perspective: "The more numerous the viewpoints," he says, "the less conclusive the reading. One had better conclude that the painting is not done in perspective at all -- something hardly surprising with regard to most other Manets" (p.000). But perspectival painting has always been this way. Any perspectival picture that has more than a single object will suffer from internal inconsistencies because every painter assembles parts that don't belong together. That is true even of careful, analytically minded paintings: Paolo Uccello's The Flood has a number of mutually inconsistent constructions, which point at different vanishing points. They have also been analysed with the help of overly precise diagrams (I have drawn a few myself), and they also focus viewers' attention on optics or Alberti and a whole range of issues associated with his name.5 Perspective pictures are endemically inconsistent and error prone. So it's historically odd to try to make everything important to fit, or to say that a rotation mirror is the most "economical" explanation "for the inconsistencies of perspective" (p. 000). Leaving aside the veracity of the rotating mirror for a moment, perpective is by nature uneconomical; it is profligate and illogical in its methods, and unreliable and inacurate in its explanations. In a domain where everything is twisted out of logical shape--and with a painter notoriously uninterested in geometry--the search for economy is itself suspect. (Nor is it immediately apparent how a rotating mirror counts as economical. To me it looks like an elaborate artifice, much less economical than Conger's rudimentary lines and ellipses, which a painter can draw almost without thinking.)

Complex diagrams like Conger's, and like some I have drawn, are often overinterpretations, just as de Duve says. But what is it, exactly, that sanctions the move from the contemplatin of individual constructions--the mutually inconsistent building blocks of any perspective picture--to the exclusive interest in a few either/ or propositions? Could it be something other than perspective? And possibly even something that needs perspective to be out of the way before it can even get started?

1. Counting length of text, not actual page numbers, and omitting quotations from other books, which occupy another page. See Fried, Manet's Modernism, or, The Face of Painting in the 1860s (Chicago, 1996), pp. 286-88 and 345-46.

2. For the ternary relation, see ibid., esp. p. 343.

3. See T. J. Clark, The Painting of Modern Life: Paris in the Art of Manet and His Followers (New York, 1985).

4. The two horizontal lines that are tangent to the upper rim of the bowl. The orthogonals receding to the vanishing point are arbitrary, and the lower foreshortened rectangle is entirely unfounded--it's engineered to make the construction look like an object on the table. (Why not use the ellipse formed by the base?) I am not claiming Conger is necessarly wrong: I like his guess, but it's nothing more than that.

5. Alberti is a more apposite reference here, since in Manet's time he had not achieved his canonical position or been so firmly associated with a specific kind of perspectival construction. For the inevitability of perspectival inconsistencies and errors, see my The Poetics of Perspective (Ithica, N.Y., 1994), pp. 219-34, and for a more detailed analysis of The Flood, my "Uccello, Duchamp: The End of Wit," Zeitschrift fur Asthetik ind allgemeine Kunstwissenschaft 36 (1993): 199-224.

James Elkins is associate proffesor of art history, theory, and criticism at the School of the Art Institute of Chicago. His books include The Object Stares Back: On the Nature of Seeing (1996) and Our Beautiful, Dry, and Distant Texts: Art history as Writing (1997).